1. Technical Field
The present disclosure concerns in general resonant switching converters circuits and in particular a control method of a resonant dc-dc converter aimed to optimize conversion efficiency (i.e., the ratio between the power provided to the load and that drawn from the input source) at low load, and a circuital implementation thereof, preferably realized in integrated form.
2. Description of the Related Art
Resonant converters represent a broad class of switching converters and include a resonant circuit playing an active role in determining the input-output power flow. In these converters, a bridge (half-bridge) consisting of four (or two) power switches (typically power MOSFETs) supplied by a dc voltage generates a square voltage wave that is applied to a resonant circuit (also termed resonant tank) tuned to a frequency close to the fundamental frequency of the square wave. Because of its selective response, the resonant circuit mainly responds to the fundamental component and negligibly to the higher order harmonics of the square wave. As a result, the circulating power may be modulated by varying the frequency of the square wave, holding the duty cycle constant at 50%. Moreover, depending on the resonant circuit configuration, the currents and/or voltages associated with the power flow have a sinusoidal or piecewise sinusoidal shape.
These voltages and/or currents are rectified and filtered so as to provide DC power to the load. In offline applications (i.e., those operated from the power line), the rectification and filtering system supplying the load is coupled to the resonant tank circuit by means of a transformer providing galvanic isolation between the source and the load, to comply with safety regulations. As in every isolated dc-dc converters, also in this case a distinction is made between a primary side (as related to the primary winding of the transformer) connected to the input source and a secondary side (as related to the secondary winding(s) of the transformer) providing power to the load through the rectification and filtering system.
As an example of resonant converter, FIG. 1 shows the so-called LLC resonant converter, probably today's most widely used resonant converter, especially in its half-bridge version. The designation LLC stems from the fact that the resonant tank employs two inductors (L) and a capacitor (C).
The resonant converter comprises a “totem-pole” of transistors M1 and M2 connected between the input voltage source node Vin and ground GND, controlled by a control circuit. The common terminal HB between the transistors M1 and M2 is connected to a resonant tank comprising a series of a capacitor Cr, an inductance Ls and another inductance Lp connected in parallel to a transformer with a center-tap secondary winding. The two windings of the center-tap secondary are connected to the anodes of two diodes D1 and D2, whose cathodes are both connected to the parallel of a capacitor Cout and a resistance Rout; the output voltage Vout of the resonant converter is across said parallel while the DC output current Iout flows through Rout.
Resonant converters offer considerable advantages as compared to traditional switching converters (which are not resonant, but typically PWM—Pulse Width Modulation—controlled): waveforms without steep edges, low switching losses in the power switches due to their soft-switching operation, high conversion efficiency (>95% is easily reachable), ability to operate at high frequencies, low EMI generation (Electro-Magnetic Interference). All these features make resonant converters ideal candidates when high power density is to be achieved, that is, when conversion systems capable of handling considerable power levels in a relatively small space are preferred.
As in most DC-DC converters, the output voltage is kept constant against changes in the operating conditions (i.e., the input voltage Vin and the output current lout) through a control system that uses closed-loop negative feedback. As shown in the block diagram of FIG. 2, this is achieved by comparing a portion of the output voltage Vout to a reference voltage Vref, their difference (error signal) is amplified by an error amplifier whose output Vc (control voltage) is transferred to the primary side across the isolation boundary typically via an optocoupler. The optocoupler changes the control voltage Vc into a control current IFB. Note that normally the circuit arrangement comprising the error amplifier and the optocoupler is such that the control voltage Vc and the control current IFB change in opposite directions: if Vc increases IFB decreases, if Vc decreases, IFB increases. The control current IFB modifies a quantity X within the converter which the power carried by the converter substantially depends on.
In resonant converters, as mentioned earlier, this significant quantity is the switching frequency of the square wave stimulating the resonant tank (X=ƒsw). In nearly all practical resonant converters, if frequency rises the delivered power decreases and vice versa.
A consideration common to many applications of switching converters, resonant and not, is that conversion efficiency is maximized also under light load conditions to comply with regulations and recommendations on energy saving (e.g., EnergyStar, CEC, Eu CoC, Climate Savers, etc.).
A popular technique for optimizing light load efficiency in all switching converters (resonant and not) is to make them work in the so-called “burst-mode”. With this operating mode the converter works intermittently, with series (bursts) of switching cycles separated by time intervals during which the converter does not switch (idle time). When the load is such that the converter has just entered burst-mode operation, the idle time is short; as the load decreases, the duration of the bursts decreases as well and the idle time increases. In this way, the average switching frequency is considerably reduced and, consequently, so is the effect of the two major contributors to power losses at light load:
1) switching losses associated to the parasitic elements in the converter
2) conduction losses related to the flow of reactive current in the resonant tank (e.g., the magnetizing current in the transformer). In fact, this current only flows while the converter is switching and is essentially zero during the idle time.
The duration of the bursts and the idle time are determined by the feedback loop so that the output voltage of the converter always remains under control. To explain the mechanism governing this operation it is convenient to refer to a concrete example.
FIG. 3 shows how burst-mode operation is implemented in the integrated control circuit L6599 by STMicroelectronics, as well as a simplified schematic of its internal current-controlled oscillator (CCO). FIG. 4 shows the oscillator waveform of the CCO, its relationship with the gate drive signals for M1 and M2 produced by the pulse-train generator and the voltage of the half-bridge midpoint HB, i.e., the square wave voltage applied to the resonant tank.
The CCO is programmed by means of the capacitor C1 connected from pin CF to ground and by the current IR sourced by the pin RFmin, which provides an accurate reference voltage Vr (=2 V). IR is internally mirrored and a current KM·IR is alternately sourced and sunk from pin CF, originating a symmetrical triangular waveform included between a peak value (=3.9 V) and a valley value (=0.9 V) across C1. As a result, the higher the current IR, the faster C1 is charged and discharged and the higher the oscillation and switching frequency (ƒosc). Denoting with ΔVosc, the peak-to-valley swing of the oscillator (=3 V), the following relationship can be found:
The current IR is the sum of the current flowing through R1 (=Vr/R1) and the current IFB sunk by the phototransistor of the optocoupler OC that transfers the control voltage Vc across the isolation boundary. Therefore, the current IFB actually modulates IR, closing the feedback loop that regulates the output voltage of the converter and making it work at a frequency given by: